Abstract
When we are dealing with online dynamical systems, i.e. Adaptive-Network-Based Fuzzy Inference Systems (ANFIS), one of the major problems is the time used into the learning process. In this way, Backpropagation algorithm is used for training the premise parameters in a hybrid learning procedure. However, it depends on parameters that must be tuned heuristically. This provides that, in some way, training is not useful completely. Nevertheless, fuzzy logic can be put into practice and assume parameters involved in the backpropagation algorithm could be treated as fuzzy elements. In this paper, we implement fuzzy logic controllers inside the backpropagation algorithm and expand this model for training Neural Networks and ANFIS to achieve convergence in short periods of time in order to decrease the time process on-line. At this point, LabVIEW will be used as platform to validate this approach.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have